A) Verification of Flag-HA-ATF5 expression from the pCIN4-Flag-HA-ATF5 construct. In vitro translation (Promega) analysis of the pCIN4-Flag-HA-ATF5 showed that a 35 kd protein, consistent with the predicted size of double-tagged Flag-HA-ATF5, was produced as expected. pCIN4-Flag-HA-p53 (Positive Control) and pCIN4 (negative control) were used. Molecular markers (kd) are labeled on the right side. B) Selection of a C6 cell line that stably expresses Flag-HA-ATF5. pCIN4-Flag-HA-ATF5 or pCIN4 were transfected into C6 glioma cells and stable transfected cell lines were selected in growth medium containing G418 (0.8 mg/ml). Cell extracts were prepared from C6-pCIN4 or C6-pCIN4-Flag-HA-ATF5 cells and an equal amount (50 μg) of cell extracts from indicated cells was loaded for each lane and resolved by 10% SDS-PAGE. Immunoblotting (IB) analysis was carried out using indicated antibodies. Solid triangles indicate expression of Flag-HA-ATF5 (35 kd); open triangle indicates endogenously expressed ATF5 that is 22 kd (5). The difference in size (about 13 kd) between the ectopically expressed Flag-HA-ATF5 and the endogenous ATF5 is contributed by the Flag-HA-tag as well as by the apparent preference of the second potential in-frame Kozak start site that is used to produce the endogenous ATF5 in every type of cell tested so far (2, 5). The two translation start sites are separated by 100 amino acids, or about 11 kd. Molecular weights (kd) are labeled on the left side. C) Analysis of ATF5-containing complexes isolated by a dual-tag/double-immunoprecipitation (IP) strategy. Whole cell extracts from C6-pCIN4 and C6-pCIN4-Flag-HA-ATF5 cells were subjected to affinity chromatography on M2 (Flag antibody) agarose beads and an HA-affinity column. The bound proteins were eluted and examined for presence of Flag-HA-ATF5 bait, as described in Materials and Methods. A fraction of this eluate was re-IPed with anti-HA beads and subjected to 10% SDS-PAGE, followed with IB using an anti-Flag antibody. The Flag-HA-ATF5 in the elute gave a prominent band at 35 kd on the Western immunoblot.

A) EMSA assays on the ATF5 consensus DNA binding sequence and extracts of C6 cells stably transfected with pCIN4 or pCIN4-Flag-HA-ATF5. While probe alone (biotin-labeled ATF5 consensus DNA binding sequence; lane 1) or probe incubated with anti-HA (lane 2) or anti-ATF5 (lane 6) anti-bodies did not show any binding, a specific ATF5 consensus DNA-binding activity was seen in extracts of C6-pCIN4 (lane 3) and of C6-pCIN4-Flag-HA-ATF5 (lane 7) cells. The stronger signal of the shifted band in lane 7 indicated additional binding by over-expressed ATF5. Addition of an anti-HA antibody did not affect binding of endogenous ATF5 to the probe (comparing lanes 3 and 4). In contrast, addition of the anti-HA antibody reduced ATF5 binding activity in the C6-pCIN4-Flag-HA-ATF5 cells (comparing lanes 7 and 8). The ATF5 binding activity that was insensitive to anti-HA antibody was at a similar level as that in the C6-pCIN4 extract (comparing lanes 3 and 8), suggesting that the anti-HA antibody specifically abolished binding of the ATF5 consensus sequence by the over-expressed Flag-HA-ATF5. Anti-ATF5 antibody completely abolished binding in C6-pCIN4-Flag-HA-ATF5 cells (lane 5). The image was generated by aligning together two representative gels. B) Comparison of competition for ATF5 binding by ATF5CON and CRE. EMSAs were carried out as in A) using ATF5 consensus DNA binding sequence as probe and extract from C6-pCIN4-Flag-HA-ATF5 cells, except that excess of non-labeled ATF5CON (CCTCTTCCTTA) or CRE (CGTGACGTCATAG) were added to the reactions prior the addition of cell extracts. C) Mutation of the ATF5CON abolishes ATF5 binding. EMSA was performed as in A) with indicated probes. ATF5CON-Mut1 (CCGATTCCTTA; lane 2) and ATF5CON-Mut2 (CCTCTTCGGTA; lane 3) contained mutations that are located at the invariably conserved T3C4 and T9 residues in the ATF5CON (Fig 2), respectively.

A and B) Serum withdrawal promoted apoptotic death of C6 cells. C6 cells (C6) or a C6 cell line stably expressing Flag-HA-ATF5 (C6-Flag-HA-ATF5; described in Fig 1) were seeded in 24-well plates at a density of 1.5×10⁵ per well and fed with growth medium containing 10% FBS. The medium was replaced with serum-free medium the next day (Day 0) and the cells were subsequently incubated for 3 more days. Cell viability was assessed by Trypan blue staining on each day after serum withdrawal (A) while the ratio of apoptotic cells were determined by TUNEL assay (B). The numbers of surviving cells before serum withdrawal were set at 100% in A). Data are means +/- SEM (n=3). C) Downregulation of ATF5CON activity by serum withdrawal was reversed by expression of ATF5 in C6 cell. C6 cells were transfected with Control vector (pLeGFP) or ATF5 (pLeGFP-Flag-ATF5) together with the ATF5CON-luc reporter and Renilla. 24 h later, cells were either re-fed with growth medium (+serum) or subjected to serum withdrawal (-serum). Cells were harvested 24 h later and luciferase activity was determined as in Fig 4A. Data are means +/- SEM (n=3). D) Serum withdrawal led to down-regulation of ATF5CON activity in MCF-7 cells. Experiment was carried out as in C) except MCF-7 cells were used. Data are means +/- SEM (n=3).

A) Sequence of the degenerate oligonucleotide used for CASTing. N20 stands for the 20 degenerate nucleotides. 5′- and 3′-primers corresponding to the sequences flanking the N20 were used in PCR reactions to amplify the selected DNA that is bound to ATF5 complexes and immunoprecipitated between each cycle. B) Alignment of 29 sequences that were selected following six rounds of protein-DNA binding/immunoprecipitation/PCR amplification procedures (see Materials and Methods). The sequences of all 29 nucleotides that originated from “N20” and that are flanked by the primers are shown. The sequences are aligned to overlap the conserved 11 bp region that are in boldface and boxed. The lack of similarity outside the 11 bp consensus region indicated that each clone was an independent isolation event. “N”s represent undetermined nucleotides outside of the consensus region. C) ATF5 consensus DNA binding sequence. Upper Panel: the number of times each nucleotide was found in positions relative to the consensus sequence [C(C/T)TCT(T/C)CCTTA] are shown. Lower Panel: percentage conservation is calculated for each of the positions for the consensus. The high degree of conservation is particularly striking for the 9 nucleotides in the middle, omitting the first C (C1) and the last A (A11). Nucleotides T3, C4, and T9 are 100% conserved.

A) Expression of ATF5 stimulates a luciferase reporter containing the ATF5 consensus DNA binding site (ATF5CON). C6 cells were transfected with pGL3 vector (control) or pGL3-ATF5CON (ATF5CON) luciferase reporter with a construct (300ng) empty (control) or expressing ATF5 (ATF5). Renilla was used as an internal control for each transfection. 48 h after transfection, cells were harvested for determination of luciferase and renilla activities. Results are reported in relative light units and normalized with Renilla activity. Data are means +/- SEM (n=3). Insert: Western immunoblotting analysis on cell extracts made from C6 cells transfected as in A). Immunoblotting was carried out using anti-ATF5 (upper panel) or anti-β-actin (lower panel) as indicated. B) Regulation of ATF5CON in MCF-7 cells. Transfection of indicated constructs and luciferase analyses were performed as in A) except MCF-7 cells were used. Data are means +/- SEM (n=3). * Significance comparing with vector p<0.01. C) Specific activation of ATF5CON-luc by ATF5. ATF5CON-luc reporter, Renilla, and indicated mammalian expression vector expressing ATF5 and CREB or vector control were transfected into C6 cells. Determination of luciferase activities was carried out as in A). Data are means +/- SEM (n=3). D) Transfection and luciferase activity assays were carried out as in C) except a CRE(c-fos)-luciferase reporter and indicated expression constructs were used. Data are means +/- SEM (n=3).

A) Concomitant regulation of ATF5 and Egr-1 in C6 and MCF-7 cells. Total RNA was harvested from C6, C6-pCIN4-Flag-HA-ATF5 (C6-FH-ATF5), and C6-pCIN4 (C6-pCIN4) cells, and from MCF-7 cells grown either in 10% or 0% FBS for 48 hours. mRNA was reversely transcribed into cDNA using oligo(dT)12∼18. Relative levels of ATF5 and Egr-1 transcripts were determined by semi-quantitative RT-PCR using primers specific to their respective mRNAs. β-actin (Actin) was used as loading control. B) Schematic illustration of the rat Egr-1 promoter region and the wild type (WT) and mutant (Mut) Egr-1 promoter luciferase reporters. The rat Egr-1 regulatory region contains many different functional elements, including two ATF5CON sites beginning at nucleotide -1675 and -1540, respectively. P1 and P2 represent the primers flanking the ATF5CON site, which were used to amplify the 201 bp fragment of Egr-1 promoter region from -1689 to -1489; P3 and P4 represent the primers that were used to amplify a 194 bp fragment in the coding region of Egr-1 from +1137 to +1330. The sequence of the two putative ATF5 binding sites was given in the two boxes aligned with the -1675 and -1540 regions. A single tolerable mismatch (A) in the first homologous region is underlined. The wild type (WT) and mutant (Mut) Egr-1 promoter luciferase reporters are aligned at the bottom. C) ATF5 is specifically associated with the endogenous Egr-1 promoter region. ChIP assays were carried out using a kit from Upstate following manufacturer's instruction. Chromatin was prepared using C6-pCIN4-Flag-HA-ATF5 (lanes 2-7) and C6-pCIN4 (lanes 9-12) cells. The amount of chromatin used for “Input” in lanes 2, 4, 9 and 11 was 10% of that used for IP experiments. Primers 1 and 2 were used for detection of precipitated promoter (Promoter, lanes 4-7 and 11-12). Primers 3 and 4 were used to detect Egr-1 coding region (Coding, lanes 2-3 and 9-10) where no sequence resembling the ATF5CON is located, which was used as control. The Flag antibody was expected to pull down the ectopically expressed Flag-HA-ATF5 in C6-pCIN4-Flag-HA-ATF5 cells while an irrelevant X-press (Irr) antibody and a mock experiment (No Ab) were used as controls. The same Flag antibody did not pull down the Egr-1 promoter in C6-pCIN4 cells (lanes 10 and 12). D) Egr-1 promoter is activated by ATF5 and repressed by d/nATF5. Luciferase reporter assay was carried out as in Fig 4A except wild type (WT) and mutant (Mut) Egr-1 promoter-luciferase reporters and indicated ATF5 constructs were used. Data are means +/- SEM (n=3). Significance comparing with control: * p<0.01; ** p<0.05.